Re: [PATCH] mfd: rk808: Always use poweroff when requested

[Soeren Moch]

On 11.12.19 03:00, Anand Moon wrote:
> hi Soeren,
>
> On Tue, 10 Dec 2019 at 22:10, Soeren Moch <smoch@xxxxxx> wrote:
>>
>>
>> On 10.12.19 13:55, Markus Reichl wrote:
>>> Hi Anand,
>>>
>>> Am 10.12.19 um 13:42 schrieb Anand Moon:
>>>> Hi Soeren,
>>>>
>>>> On Mon, 9 Dec 2019 at 17:28, Soeren Moch <smoch@xxxxxx> wrote:
>>>>> With the device tree property "rockchip,system-power-controller" we
>>>>> explicitly request to use this PMIC to power off the system. So always
>>>>> register our poweroff function, even if some other handler (probably
>>>>> PSCI poweroff) was registered before.
>>>>>
>>>>> Signed-off-by: Soeren Moch <smoch@xxxxxx>
>>>>> ---
>>>>> Cc: Lee Jones <lee.jones@xxxxxxxxxx>
>>>>> Cc: Heiko Stuebner <heiko@xxxxxxxxx>
>>>>> Cc: linux-arm-kernel@xxxxxxxxxxxxxxxxxxx
>>>>> Cc: linux-rockchip@xxxxxxxxxxxxxxxxxxx
>>>>> Cc: linux-kernel@xxxxxxxxxxxxxxx
>>>>> ---
>>>>>  drivers/mfd/rk808.c | 11 ++---------
>>>>>  1 file changed, 2 insertions(+), 9 deletions(-)
>>>>>
>>>>> diff --git a/drivers/mfd/rk808.c b/drivers/mfd/rk808.c
>>>>> index a69a6742ecdc..616e44e7ef98 100644
>>>>> --- a/drivers/mfd/rk808.c
>>>>> +++ b/drivers/mfd/rk808.c
>>>>> @@ -550,7 +550,7 @@ static int rk808_probe(struct i2c_client *client,
>>>>>         const struct mfd_cell *cells;
>>>>>         int nr_pre_init_regs;
>>>>>         int nr_cells;
>>>>> -       int pm_off = 0, msb, lsb;
>>>>> +       int msb, lsb;
>>>>>         unsigned char pmic_id_msb, pmic_id_lsb;
>>>>>         int ret;
>>>>>         int i;
>>>>> @@ -674,16 +674,9 @@ static int rk808_probe(struct i2c_client *client,
>>>>>                 goto err_irq;
>>>>>         }
>>>>>
>>>>> -       pm_off = of_property_read_bool(np,
>>>>> -                               "rockchip,system-power-controller");
>>>>> -       if (pm_off && !pm_power_off) {
>>>>> +       if (of_property_read_bool(np,
>>>>> "rockchip,system-power-controller")) {
>>>>>                 rk808_i2c_client = client;
>>>>>                 pm_power_off = rk808->pm_pwroff_fn;
>>>>> -       }
>>>>> -
>>>>> -       if (pm_off && !pm_power_off_prepare) {
>>>>> -               if (!rk808_i2c_client)
>>>>> -                       rk808_i2c_client = client;
>>>>>                 pm_power_off_prepare = rk808->pm_pwroff_prep_fn;
>>>>>         }
>>>>>
>>>> I gave this a try on my Rock960 and Odroid N1
>>>> both got kernel panic below.
>>> I see the same on rk3399-roc-pc.
>> This is no panic, it's a harmless warning.
> Ok but my device do not come up cleanly after that, it get stuck in
> u-boot in next boot. 
I do not know Rock960 and Odroid N1, so unfortunately I cannot debug
your u-boot problem on these boards. From what you wrote the poweroff
apparently works with this patch, I have no idea what could be a
problem. After a complete power-off I would expect a clean cold boot.
And that 's what I see on my RockPro64 board.

This patch is about doing a PMIC poweroff when this method is requested
in the devicetree. If this method does not work for your boards, you
probably should remove the "rockchip,system-power-controller" property
and use PSCI poweroff or whatever is desired on your boards. This patch
does not change _how_ PMIC poweroff is done, only _that_ this method is
used when explicitly requested.
>
>> The i2c core nowadays expects a specially marked i2c transfer function
>> late in the powerdown cycle:
> You can look into similar commit.
> d785334a0d5deff30a487c74324b842d2179553d (mfd: s2mps11: Add manual
> shutdown method for Odroid XU3)
I cannot see what should be similar in this patch. This patch is about a
totally different PMIC and how this needs to be programmed to work
properly on another different board.

Soeren
>
>> diff --git a/drivers/i2c/busses/i2c-rk3x.c b/drivers/i2c/busses/i2c-rk3x.c
>> index 1a33007b03e9..cec115e0afa4 100644
>> --- a/drivers/i2c/busses/i2c-rk3x.c
>> +++ b/drivers/i2c/busses/i2c-rk3x.c
>> @@ -1126,6 +1126,7 @@ static u32 rk3x_i2c_func(struct i2c_adapter *adap)
>>
>>  static const struct i2c_algorithm rk3x_i2c_algorithm = {
>>      .master_xfer        = rk3x_i2c_xfer,
>> +    .master_xfer_atomic    = rk3x_i2c_xfer, /* usable for PMIC poweroff */
>>      .functionality        = rk3x_i2c_func,
>>  };
>>
>> ---
>> It is only used for powerdown. The regular i2c xfer function works.
>>
>> Heiko, should I send a formal patch for that?
>>
>> Soeren
>>
>>>> [   58.305868] xhci-hcd xhci-hcd.0.auto: USB bus 5 deregistered
>>>> [   58.306747] reboot: Power down
>>>> [   58.307106] ------------[ cut here ]------------
>>>> [   58.307510] No atomic I2C transfer handler for 'i2c-0'
>>>> [   58.308007] WARNING: CPU: 0 PID: 1 at drivers/i2c/i2c-core.h:41
>>>> i2c_transfer+0xe4/0xf8
>>>> [   58.308696] Modules linked in: snd_soc_hdmi_codec dw_hdmi_i2s_audio
>>>> rockchipdrm analogix_dp brcmfmac nvme dw_mipi_dsi nvme_core dw_hdmi
>>>> panfrost cec brcmutil drm_kms_helper gpu_sched cfg80211 hci_uart drm
>>>> btbcm crct10dif_ce snd_soc_simple_card bluetooth snd_soc_rockchip_i2s
>>>> snd_soc_simple_card_utils snd_soc_rockchip_pcm phy_rockchip_pcie
>>>> ecdh_generic rtc_rk808 ecc pcie_rockchip_host rfkill rockchip_thermal
>>>> ip_tables x_tables ipv6 nf_defrag_ipv6
>>>> [   58.312150] CPU: 0 PID: 1 Comm: shutdown Not tainted
>>>> 5.5.0-rc1-dirty #1
>>>> [   58.312725] Hardware name: 96boards Rock960 (DT)
>>>> [   58.313131] pstate: 60000085 (nZCv daIf -PAN -UAO)
>>>> [   58.313551] pc : i2c_transfer+0xe4/0xf8
>>>> [   58.313889] lr : i2c_transfer+0xe4/0xf8
>>>> [   58.314225] sp : ffff80001004bb00
>>>> [   58.314516] x29: ffff80001004bb00 x28: ffff00007d208000
>>>> [   58.314981] x27: 0000000000000000 x26: 0000000000000000
>>>> [   58.315446] x25: 0000000000000000 x24: 0000000000000008
>>>> [   58.315910] x23: 0000000000000000 x22: ffff80001004bc74
>>>> [   58.316375] x21: 0000000000000002 x20: ffff80001004bb58
>>>> [   58.316841] x19: ffff0000784f0880 x18: 0000000000000010
>>>> [   58.317305] x17: 0000000000000001 x16: 0000000000000019
>>>> [   58.317770] x15: ffffffffffffffff x14: ffff8000118398c8
>>>> [   58.318236] x13: ffff80009004b867 x12: ffff80001004b86f
>>>> [   58.318701] x11: ffff800011851000 x10: ffff80001004b7f0
>>>> [   58.319166] x9 : 00000000ffffffd0 x8 : ffff800010699ad8
>>>> [   58.319631] x7 : 0000000000000265 x6 : ffff800011a20be9
>>>> [   58.320096] x5 : 0000000000000000 x4 : 0000000000000000
>>>> [   58.320561] x3 : 00000000ffffffff x2 : ffff800011851ab8
>>>> [   58.321026] x1 : d375c0d4f4751f00 x0 : 0000000000000000
>>>> [   58.321491] Call trace:
>>>> [   58.321710]  i2c_transfer+0xe4/0xf8
>>>> [   58.322020]  regmap_i2c_read+0x5c/0x98
>>>> [   58.322350]  _regmap_raw_read+0xcc/0x138
>>>> [   58.322694]  _regmap_bus_read+0x3c/0x70
>>>> [   58.323034]  _regmap_read+0x60/0xe0
>>>> [   58.323341]  _regmap_update_bits+0xc8/0x108
>>>> [   58.323707]  regmap_update_bits_base+0x60/0x90
>>>> [   58.324099]  rk808_device_shutdown+0x38/0x50
>>>> [   58.324476]  machine_power_off+0x24/0x30
>>>> [   58.324823]  kernel_power_off+0x64/0x70
>>>> [   58.325159]  __do_sys_reboot+0x15c/0x240
>>>> [   58.325504]  __arm64_sys_reboot+0x20/0x28
>>>> [   58.325858]  el0_svc_common.constprop.2+0x88/0x150
>>>> [   58.326279]  el0_svc_handler+0x20/0x80
>>>> [   58.326607]  el0_sync_handler+0x118/0x188
>>>> [   58.326960]  el0_sync+0x140/0x180
>>>> [   58.327251] ---[ end trace b1de39d03d724d01 ]---
>>>>
>>>> -Anand
>>>>
>>>> _______________________________________________
>>>> Linux-rockchip mailing list
>>>> Linux-rockchip@xxxxxxxxxxxxxxxxxxx
>>>> http://lists.infradead.org/mailman/listinfo/linux-rockchip
>>>>
>>> GruÃ,
> -Anand